Polymorphism in natural alleles of the avirulence gene Avr1c is associated with the host adaptation of Phytophthora sojae

Abstract

Phytophthora sojae is a destructive pathogen of soybean that is widely distributed in the world. The interaction between P. sojae and soybean follows the gene-for-gene model. The use of resistant soybean cultivars is the primary and most effective method to combat the disease. However, variation in the Avr genes of the pathogen enables it to evade host defenses. We collected 81 isolates from four major soybean-production areas in China to analyze the polymorphism of Avr genes in P. sojae field population. The virulence of these isolates towards 14 differential soybean lines indicated complex pathotypes in P. sojae field population in China. In this study we found that Rps1c, which is cognate with Avr1c, could be deployed in Heilongjiang, Shandong, and Jiangsu Provinces but not in Anhui Province. To determine the mechanism by which Avr1c escapes recognition by Rps1c, we analyzed the polymorphism of Avr1c gene in 50 isolates of a field population of P. sojae and found multiple novel genotypes related to virulence and avirulence. By performing infection assays and gene co-bombardment, we showed that the K105 amino-acid residue was under strong positive selection and was a determinant of the avirulence of Avr1c. Structural analysis showed that K105 was exposed on the surface of the protein, suggesting it to be a critical site for interacting with Rps genes or their associated proteins.

Background

Phytophthora root and stem rot caused by Phytophthora sojae (M. J. Kaufmann & J. W. Gerdemann) is a destructive disease of soybean. Root rot of soybean caused by P. sojae was first observed in Indiana in 1948 and then in Ohio in 1951 (Schmitthenner 1985), which can result in soybean seedlings damping off and cause an annual loss of approximately $1–2 billion on soybean production worldwide (Tyler 2007).

The use of resistant soybean cultivars is the primary and most effective method to manipulate the disease (Sugimoto et al. 2012). P. sojae and soybean follows the gene for gene model and using resistant cultivars controlled by a single resistance (Rps) gene to combat the disease can result in complete resistance (Schmitthenner 1985). More than 28 Rps genes were identified so far, including Rps1a to Rps12, RpsYD25, RpsYu25, RpsYD29, RpsUN1, RpsJS, RpsSN10, RpsYB30, RpsWaseshiroge and RpsZS18 (Sun et al. 2014; Sahoo et al. 2017), and only 4 genes encoding nucleotide binding-leucine rich repeat (NB-LRR) proteins were isolated from the Rps1k locus (Gao et al. 2005).

However, avirulence (Avr) genes are under strong selective pressure in P. sojae field population, which makes the resistance conferred by Rps genes to be overcome by P. sojae in 8–15 years (Dorrance et al. 2016). Nine Avr genes of P. sojae have been cloned to date: Avr1a (cognate Rps gene: Rps1a) (Qutob et al. 2009), Avr1b (Rps1b) (Shan et al. 2004), Avr1c (Rps1c) (Na et al. 2014), Avr1d (Rps1d) (Yin et al. 2013), Avr1k (Rps1k) (Song et al. 2013), Avr3a/5 (Rps3a, Rps5) (Qutob et al. 2009), Avr3b (Rps3b) (Dong et al. 2011), Avr3c (Rps3c) (Dong et al. 2009), and Avr4/6 (Rps4, Rps6) (Dou et al. 2010). Avr1b-1 was the first Avr gene of P. sojae to be cloned. Point mutations and loss of transcript of Avr1b-1 induced virulence (Shan et al. 2004). In addition, the transposable elements insertion resulted in deletion of Avr1b-1 from Chinese isolates (Cui et al. 2012). Avr1d is present in P. sojae strains avirulent to Rps1d, but is absent from the genome of virulent strains (Na et al. 2013). The virulent alleles of Avr4/6 compared with its avirulent ones have nucleotide substitutions and deletions in the 5’untranslated region but not in the protein-coding region (Dou et al. 2010). The transgenerational silencing of Avr3a by 25-nucleotide RNA molecules enhances the virulence of Rps3a (Qutob et al. 2013).

Rps1c is the most commonly used Rps gene in commercial soybean cultivars in the United States (Slaminko et al. 2010) and in domestic cultivars in China (unpublished data). However, the pathogens were under high selection pressure, causing complex pathotypes in yield population so that the resistant cultivars controlled by a single Rps gene were easy to be overcome (Dorrance et al. 2016). So monitoring of a pathogen population’s adaptation to resistance (R) genes is essential for long-term management of plant diseases. In our work, we found that Rps1c could be deployed in most, but not all, soybean-production areas in China, indicating that Avr1c is under strong selective pressure. Deletions and silencing, but not point mutations, in the cognate avirulence gene Avr1c of P. sojae result in its escape from recognition by Rps1c (Na et al. 2014; Arsenault-Labrecque et al. 2018). To determine the mechanism by which Avr1c escapes recognition by Rps1c, we collected P. sojae isolates from major soybean-production areas in China and analyzed them together with several isolates stored in our laboratory.

Results

Virulence proportion of P. sojae population collected in China

To study the pathotypes of P. sojae in China, we collected 81 isolates from diseased plants and soil samples in 4 major soybean-production regions, including Heilongjiang at northeast China (36 isolates), Shandong (13 isolates), Anhui (19 isolates) and Jiangsu (13 isolates) Provinces at central-east China, during 2015–2016. All of these isolates were recovered by single zoospore and evaluated for pathotypes on 14 soybean differential lines with Williams as a blank control. Of the 81 isolates, more than 60% were virulent on soybean lines carrying Rps5, Rps1d, Rps3c, Rps7 or Rps8, 21% were virulent on Rps1k, 22% were virulent on Rps1c, and 33% were virulent on Rps1a and Rps3a (Fig. 1a). These results indicated that resistance conferred by Rps1k, Rps1c and Rps1a are less likely to be overcome by these P. sojae isolates. In addition, 74 pathotypes were identified in these 81 isolates (Additional file 1: Table S1), with seven pathotypes isolated twice while the rest of pathotypes only once, showing complex pathotypes in Chinese P. sojae population. The isolate PsSD6 was virulent on only Rps1a gene, six isolates were virulent on two Rps genes, and approximately 75% of isolates were virulent on six or more Rps genes.

Fig. 1

The virulence proportion of Phytophthora sojae isolates on 14 differential lines collected in China. a The virulence proportion of all 81 isolates and b Proportion of isolates collected in Heilongjiang, Anhui, Shandong, and Jiangsu Province on 14 differential lines. Every assay was repeated independently twice. The x-axis indicated the differential lines containing the 14 Rps genes

Of the 36 isolates obtained from Heilongjiang Province, less than 9% were virulent on Rps1b, Rps1c and Rps6 while 19%, 22% and 27% were virulent on Rps1k, Rps3a and Rps1a, respectively. More than 50% isolates were virulent on Rps1d, Rps2, Rps3b, Rps3c, Rps5 and Rps8, indicating that those Rps genes have already been overcome by P. sojae isolates in Heilongjiang Province. Of the 13 isolates obtained from Shandong Province, only one was virulent on Rps1k and Rps3a individually. Meanwhile, three isolates were virulent on Rps1a and Rps1c. Four or more isolates were virulent on 10 other Rps genes. In the case of the isolates collected from Anhui and Jiangsu Provinces, except that 23% isolates were virulent on Rps1k in Jiangsu, more than 30% isolates were virulent on each of the remaining 13 Rps genes (Fig. 1b). The survey showed that significant differences of pathotypes existed among different areas, and it’s important to deploy cultivars carrying different Rps genes according to the virulence proportion of P. sojae.

Polymorphism of Avr1c in Chinese isolates

In this study, a P. sojae field population was collected from Heilongjiang, Shandong, Anhui and Jiangsu Provinces, the major soybean-production areas in China, where we found that 5.6%, 23.1%, 47.4% and 30.8% of the 81 isolates were virulent on Rps1c, respectively (Fig. 1). So cultivars containing Rps1c was nearly overcome in Anhui Province. To elucidate how the cognate avirulence gene Avr1c escaped recognition by Rps1c, we collected 50 (35 virulent and 15 avirulent) P. sojae isolates from China to analyze the polymorphism of Avr1c using P6497 as the control (Additional file 2: Table S2). Two pairs of primers were used to amplify the open reading frame (ORF), the upstream and downstream regions of Avr1c from genomic DNA (Additional file 3: Table S3).

Avr1c was absent in 14 virulent isolates (Fig. 2a), and 10 nucleotide sequences (Additional file 4: Figure S1) were detected in the remaining 21 virulent and 16 avirulent isolates. Alleles Avr1c-3-1 and Avr1c-3-2 encode the same amino acids (Additional file 5: Figure S2). None of the Avr1c sequences contained premature stop codons or frameshift mutations. Multiple variants of the nine predicted Avr1c amino acid sequences revealed a highly polymorphic family. There were 37 polymorphic sites among the nine amino acid genotypes (Fig. 2b, c), of which six were virulent and three were avirulent types. Among the avirulent isolates, thirteen contained the Avr1c-1 allele, two contained the Avr1c-2 allele, and one contained the Avr1c-3 allele. Eight of the twenty-four virulent isolates contained the Avr1c-1, six harbored the Avr1c-7, three contained the Avr1c-4, and one each had the Avr1c-5, Avr1c-6, Avr1c-8, and Avr1c-9 allele (Additional file 2: Table S2). We found that the absence of Avr1c alleles induced virulence in Rps1c, but the presence of Avr1c-1 alleles did not always induce avirulence in Rps1c. Moreover, the presence of Avr1c-5, Avr1c-6 and Avr1c-7 alleles induced virulence in Rps1c.

Fig. 2

Polymorphism analysis of Avr1c. a PCR amplification of Avr1c. A, the isolates were avirulent on Rps1c; V, virulent on Rps1c. P6497 was used as the control. b Amino-acid sequence of Avr1c-1. Conserved sites (grey), variation sites (blue), and key recognition site (green). The consensus sequence and sequence diversity are depicted as sequence logos. c Amino-acid sequences encoded by Avr1c variants. Identities are indicated by dots and substitutions by the substituted residue. *, stop codon

Avr1c residues were under diversifying selection

The Yn00 tool in the PAML 4.9e package (Yang 2007) was used to count the ratio of the nonsynonymous substitution (dN) to synonymous substitution (dS) to determine the selection pressures underlying sequence diversification in the 10 Avr1c nucleotide sequences. The 10 nucleotide sequences made up 45 pairwise comparisons for the test. We found that dN was greater than dS (ω = dN/dS > 1) in 42 of 45 pairwise comparisons, showing that the full-length sequences were under strong selection pressure in the Avr1c gene, particularly at the C-terminal domain (Additional file 6: Table S4). To detect Avr1c residues that were positively selected, we used the maximum likelihood method and CodeML program in the PAML 4.9e package (Yang 2007). We collected all divergent Avr1c alleles and conducted the likelihood ratio test (LRT) between the null model M7 (β distribution) and the alternative model M8 (β + ω distribution). The data suggested that the Avr1c alleles are under positive selection. The Bayes empirical Bayes analysis revealed that the six amino acid sites (94 E, 96 D, 99 K, 105 K, 113 S, 118 N) were under positive selection with confidence > 99% (Table 1).

Table 1 Likelihood ratio test results for Avr1c

Three variants in virulent isolates were recognized by Rps1c

To investigate whether those variants in virulent isolates evade recognition of Rps1c by mutation, we transiently expressed the divergent Avr1c protein fused GFP in Nicotiana benthamiana without signal peptide to ensure the expression (Additional file 7: Figure S3). Then the Avr1c variants were transiently expressed in Rps1c leaves by co-bombardment. The isolates containing Avr1c-2 and Avr1c-3 were all avirulent on Rps1c, so we speculated that Avr1c-2 and Avr1c-3 could be recognized by Rps1c. The only difference between Avr1c-8 and Avr1c-9 was a single change from K (lysine) to E (glutamic acid) at position 85 (Fig. 2c). We expressed Avr1c-1, Avr1c-4, Avr1c-5, Avr1c-6, Avr1c-7 and Avr1c-8, together with the reporter gene GUS, in leaves of Williams 79 (Rps1c). The results showed that there was no significant difference in the number of blue spots (surviving cells) in leaves expressing Avr1c-8 and Avr1c-4 compared with that of empty vector (EV), suggesting that Avr1c-8 and Avr1c-4 were not recognized by Rps1c, while there was significant difference in leaves expressing Avr1c-5, Avr1c-6, and Avr1c-7 compared with that of empty vector (EV) (Fig. 3), indicating that these three variants were recognized by Rps1c.

Fig. 3

The response of five Avr1c variants expressed in Williams (rps1c) and Williams 79 (Rps1c). Blue spots, surviving cells. Right, the ratio of GUS-positive blue spots following bombardment with Avr1c compared to EV. Bars represent standard errors from three independent replicates. **, P < 0.01 by t-test

Avr1c was silenced in some virulent Chinese isolates

The Avr1c-1, Avr1c-5, Avr1c-6 and Avr1c-7 variants were all recognized by Rps1c when transiently expressed in Rps1c leaves. We hypothesized that the virulent isolates containing those four variants escaped the recognition of Rps1c by silencing. To test the transcript level of divergent Avr1c alleles in different isolates, we collected RNA samples at 0.5, 1.5, 3, 6, 9 and 12 hours post inoculation (hpi) to conduct RT-qPCR (Additional file 8: Figure S4). The results showed that the highest expression level of Avr1c occured at 3 hpi, and then reduced at 6 hpi. Therefore, we collected samples from different isolates containing the divergent Avr1c alleles at different infection stages and merged the samples to extract RNA and to conduct RT-qPCR. The results showed that Avr1c transcript was not detected in 16 virulent isolates containing the Avr1c-1, Avr1c-5, Avr1c-6 and Avr1c-7 alleles. These data suggested that Avr1c in these Chinese P. sojae isolates escaped recognition by Rps1c by gene silencing. Transcript levels of Avr1c in isolates containing the Avr1c-4, Avr1c-8 and Avr1c-9 alleles were quantified, although the expression level of the Avr1c-4 was significantly lower than the Avr1c-8 and Avr1c-9 alleles (Fig. 4). We hypothesized that point mutations in the 10 virulent isolates with these three alleles were responsible for the escape from the recognition of Avr1c by Rps1c. This may explain why these alleles were recognized by Rps1c.

Fig. 4

Expression levels of Avr1c alleles determined by RT-qPCR. The samples at infection stages at 0.5, 1.5, 3, 6, 9, and 12 hpi were collected and mixed. The RNA of the merged samples were extracted to conduct RT-qPCR. The error bars were from three independent replicates. PsSD1 (Avr1c deleted) as the control. Nt, no transcription. A, the isolates were avirulent on Rps1c; V, virulent on Rps1c

C-terminal of Avr1c-1 is critical for recognition by Rps1c

The isolates containing Avr1c-4 and Avr1c-8 were virulent on Rps1c, and the two variants were not recognized by Rps1c when transiently expressed in Rps1c leaves. So we speculated that isolates containing Avr1c-4 and Avr1c-8 proteins escaped recognition by point mutation. Firstly, we infused Avr1c-1^1-104 with Avr1c-8^105-126 (Avr1c^1N + 8C) and Avr1c-8^1-104 with Avr1c-1^105-126 (Avr1c^8N + 1C) combining the sequence of Avr1c-4 and Avr1c-8. The Avr1c-1, Avr1c-8, Avr1c^1N + 8C and Avr1c^8N + 1C were transiently expressed in Rps1c leaves using gene co-bombardment. The number of blue spots of Avr1c^8N + 1C was significantly smaller than that of the control empty vector (EV), while the blue pots of Avr1c^1N + 8C were similar with empty vector (EV) (Fig. 5), indicating that the C-terminal of Avr1c-1 is important for recognition by Rps1c.

Fig. 5

C-terminal of Avr1c is critical for recognition by Rps1c. a The wild-type and mutants transiently expressed in rps1c and Rps1c leaves. The blue pots (surviving cells) of rps1c and Rps1c leaves transiently expressing Avr1c^1N + 8C, Avr1c^8N + 1C, and the wild-type. b Ratio of GUS-positive blue spots following bombardment with Avr1c variants and mutants compared to EV. Bars represent standard errors from three independent replicates. **, P < 0.01 by t-test

K105 is the key site of Avr1c recognized by Rps1c

The C-terminal 105–126 of Avr1c-1 was required for its recognition by Rps1c. Considering the Avr1c-5 and Avr1c-6 were all recognized by Rps1c, we speculated that the point mutations in Avr1c-5 and Avr1c-6 were all overcome. To further investigate the key amino acids that determined the recognition of Avr1c by Rps1c, we compared the Avr1c-1, Avr1c-2 and Avr1c-7, which were recognized by Rps1c, with Avr1c-4, Avr1c-8, which were not recognized by Rps1c. K (lysine) 105 and R (arginine) 124 were found to be different between the variants (Fig. 6a), indicating that the two amino acids may be important for recognition of Avr1c. We constructed the mutations Avr1c-1^K105E, Avr1c-1^K105T, Avr1c-8^T105K and Avr1c-4^E105K. All mutations were ligated into pFF19 and were transiently expressed in Rps1c leaves using gene co-bombardment. Following their bombardment into Rps1c leaves, the expression of Avr1c-8^T105K and Avr1c-4^E105K induced significantly more cell death as compared to empty vector, while Avr1c-1^K105E and Avr1c-1^K105T did not (Fig. 6b), which indicates that K105 is important for recognition of Avr1c by Rps1c. However, we expressed Avr1c-1^R124S and Avr1c-8^S124R in Rps1c leaves, and the blue pots in leaves expressing the two variants were similar to those in wild-type leaves with Avr1c-1 and Avr1c-8, respectively. There was no difference in cell death between Avr1c-1^R124S and Avr1c-8^S124R versus the wild type in Rps1c leaves, indicating that amino acid R124 was not a key point for Avr1c recognition by Rps1c (Fig. 6b). Interestingly, the key residue K105 we identified fell within the sites that were under positive selection (Table 1). Our results demonstrated that K105 is a key residue determining the recognition of Avr1c by Rps1c and K105 was under strong positive selection, suggesting that Avr1c evolved to escape from Rps1c recognition through substitution at this key residue.

Fig. 6

The key site of Avr1c recognized by Rps1c. a Avr1c^1N + 8C, Avr1c^8N + 1C, and the wild-type allele expressed in Rps1c leaves. b Sequence analysis of the key site. R, recognized by Rps1c; NR, not recognized by Rps1c. c Mutants of Avr1c-1, Avr1c-8, and Avr1c-4 expressed in Rps1c leaves. Right bar chart, frequency of GUS-positive blue spots following bombardment with Avr1c compared to EV. Bars represent standard errors from three independent replicates. **, P < 0.01 by t-test

Discussion

In this study, we tested the pathotypes of P. sojae isolates from four representative major soybean-production areas in China (Heilongjiang, Anhui, Shandong and Jiangsu Provinces), to monitor changes and shifts of P. sojae population. We found that the virulence proportion of P sojae isolates toward Rps1a, Rps1b, Rps1c, Rps1k, Rps3a and Rps6 are less than 25% in Heilongjiang, indicating that cultivars containing these resistance genes could defend most isolates in Heilongjiang Province. In line with our data, Wen et al. (2018) collected P. sojae isolates from Heilongjiang Province in three continuous years and suggested that Rps1c, Rps1k and Rps3a were more effective than other Rps genes. In addition, our results demonstrated that Rps1a, Rps1c, Rps1k and Rps3a were effective in Shandong Province, Rps1a, Rps1c and Rps1k were effective in Jiangsu Province and only Rps1k was effective in Anhui Province. Since the virulence proportion of the isolates on Rps1k in Anhui was higher than 30%, the resistance gene Rps1k was nearly about to be overcome. It is reported that virulence towards Rps1a, Rps1c and Rps1k, which were widely deployed in the USA in the last many years, increased continually, but Rps6 and Rps8 were effective against the majority of isolates collected in northern regions of the sampled area (Dorrance et al. 2016). However, our data showed that Rps6 and Rps8 were not effective in China except Heilongjiang Province. Together, our results revealed increasing pathotype complexity of P. sojae population in China, and suggested the importance of stacked Rps genes in combination with high partial resistance to limit losses caused by P. sojae in breeding programs.

It is reported that Rps1c is the most commonly used Rps gene in commercial soybean cultivars in the United States (Slaminko et al. 2010). We identified Rps genes in 41 domestic cultivars in China, 9 of which contained Rps1c, indicating that Rps1c is the most commonly used Rps gene in domestic cultivars in China (unpublished data). However, Avr genes are under strong selective pressure in field populations of P. sojae, which drives Rps genes to be overcome. We found that the virulence proportion of P. sojae population on Rps1c was relatively lower in Heilongjiang, Shandong, and Jiangsu Provinces than in Anhui Province, indicating that cultivars containing Rps1c could defend against most isolates in Heilongjiang, Shandong, and Jiangsu Provinces, but not in Anhui Province. According to the Rps genes identification, 3 of 6 domestic cultivars from Anhui Province contained Rps1c, indicating that Rps1c is the most commonly used Rps gene in Anhui (unpublished data). It is thus important to investigate the variations of Avr1c gene in Chinese populations of P. sojae. This is the first report of the variation of Avr1c in Chinese P. sojae, and also the first to show that point mutations in Avr1c are responsible for its evasion of host recognition.

The results of YN00 indicated Avr1c was under positive selection, particularly the C-terminal (Additional file 6: Table S4). BEB analysis suggested that the six residues contained in Avr1c were under strong positive selection (Yang 2007). Our further analyses demonstrated that K105 fell within the sites that were under positive selection. Pro132 is essential for both avirulence and virulence of Avr3b, and substitution of serine for glycine at position 174 in PsAvr3c resulted in evasion of Rps3c-mediated immunity (Kong et al. 2015; Huang et al. 2019). Point mutations in Avr genes enable them to evade recognition by Rps genes. The Hpa isolate Hind2 evades recognition by RPP4 due to a mutation in the functional NLS region in HaRxL103, resulting in altered subcellular localization (Asai et al. 2018). It is reported that the conserved motifs at the C-terminus of Avr are critical for their interactions with Rps genes (Dou et al. 2008a; Du et al. 2018). In this study, we also found that C-terminal of Avr1c is critical for its recognition by Rps1c. Our results indicated K105 of Avr1c was important for the Rps1c-mediated response but is not a predicted NLS using the cNLS Mapper (Kosugi et al. 2009). However, R124, which was mutated in Avr1c-8 and Avr1c-9, was not a key position. Structural simulation using SWISS-MODEL indicated that the RXLR domain comprised a bundle of three α-helices (α1, Trp80-Asn90; α2, Glu94-Ala100; and α3, Ser113-Leu123). Amino acids K105, T105 and E105 in variants were exposed on the surface of the flexible loop between the W and Y motifs. The K105 residue was mapped on the exposed surface of Avr1c (Additional file 9: Figure S5a) between the W and Y motifs (Additional file 9: Figure S5b). Also, K105 was mapped on the ribbon diagram (Additional file 9: Figure S5b–d), suggesting its importance in interactions of Avr1c with Rps genes or target proteins, although the key site was mapped on the flexible loop. Thus, K105 may be a determinant of the interaction of Avr1c with Rps or target proteins. The next step is to identify the target proteins of Avr1c.

Many Avr effectors escape recognition by cognate Rps genes in P. sojae by deletion, such as Avr1a, Avr1b, Avr1c and Avr1d (Qutob et al. 2009; Cui et al. 2012; Na et al. 2013; Na et al. 2014; Arsenault-Labrecque et al. 2018). In three isolates, Avr1c was deleted in an 8.8 kb or 10.8 kb region containing two copies of Avr1a, one copy of Avr1c, and one copy of Avh72 (Na et al. 2014; Arsenault-Labrecque et al. 2018). In our study, Avr1c was deleted in 14 of the 35 virulent isolates. We detected Avr1a and Avh72 in some, but not all, of the isolates that lacking Avr1c. Whether the entire locus or only Avr1c was deleted needs to be investigated further.

Avr1c transcript was not detected in 16 virulent isolates harboring the Avr1c-1, Avr1c-5, Avr1c-6 and Avr1c-7 alleles using the specific primers. Nucleotide substitutions in the 5′-untranslated region can silence Avr4/6 (Dou et al. 2010). In the region upstream of the Avr1c ORF (Additional file 10: Figure S6), we did not detect a silencing mutation. Arsenault-Labrecque et al. (2018) reported that the virulent isolates 5B, 5C and 45B had significantly lower expression of Avr1c compared with the avirulent isolate 28A. Furthermore, no causative mutation in Avr1c was detected, but unique mutations were present in 5B and 5C. In another study, Avr3a of P. sojae was detected in avirulent but not in virulent strains, and 25 nt RNA molecules were abundant in gene-silenced strains (Qutob et al. 2013). We hypothesized that small RNAs are responsible for silencing Avr1c, but the type of small RNA and whether the silencing was transgenerational needs further investigation.

The selective pressure exerted by R genes has driven avirulent isolates to evolve mechanisms for overcoming host resistance; e.g., deletions, point mutations or insertions, or silencing of Avr genes, which can result in complete loss of the Avr protein or the production of altered forms that do not trigger an R-gene-mediated defense response. We evaluated the variation of Avr genes and the mechanisms through which they evade recognition by cognate R genes. The ultimate aim is to engineer soybean cultivars with enhanced resistance to disease. We found that deletion, silencing and point mutations of Avr1c allow it to evade from Rps1c recognition, driving P. sojae to overcome Rps1c-mediated host defenses and providing molecular insight into the coevolution of Avr and Rps genes.

Conclusions

Phytophthora root and stem rot caused by Phytophthora sojae (M. J. Kaufmann & J. W. Gerdemann) is a destructive disease of soybean which causes approximately $1–2 billion in annual agricultural losses worldwide. Rps1c is the most commonly used resistance gene in commercial soybean cultivars in the United States and in domestic cultivars in China. However, the resistant cultivars controlled by a single Rps gene to combat the disease is easily overcome in a few years. We tested the pathotypes of P. sojae isolates from four representative major soybean-production areas in China to monitor changes and shifts of P. sojae population. Our results showed that the K105 amino-acid residue under strong positive selection is a determinant of the avirulence of Avr1c. Structural analysis showed that K105 was exposed on the surface of the protein, indicating that it to be a critical site for interacting with Rps genes or their associated proteins. We found that deletion, silencing, and point mutations of Avr1c allow it to evade from Rps1c recognition, driving P. sojae to overcome Rps1c-mediated host defenses and providing molecular insight into the coevolution of Avr and Rps genes.

Methods

Samples and P. sojae isolates collection

Soil samples and diseased plants were harvested in soybean fields from Heilongjiang, Anhui, Shandong and Jiangsu Provinces, where 100 soil samples, 43 soil samples, 45 diseased plants samples, and 68 diseased plant samples were gathered, respectively. Soil baiting for P. sojae was conducted according to the methods of Schmitthenner et al. (1994). About 10 g of each finely ground soil sample was placed uniformly in a 9 cm plastic pot bathed with 4–7 mL sterilized and deionized water, then sealed with parafilm, and inversely cultured in a greenhouse at 25 °C, with a photoperiod of 14 h light/10 h dark for 5 days. Leaf discs were put into the pot quickly after the appropriate amount of sterilized and deionized water was added to incubate for 24–48 h in the greenhouse. The baited leaf discs were washed two times with sterilized and deionized water and fostered in the greenhouse for 48 h. The zoospore suspension containing 200–300 zoospores was coated on Phytophthora selective water agar containing 50 ppm ampicillin, 50 ppm rifampicin, 50 ppm quintozene (PCNB) and 5 ppm phenamacril. The single generated zoospore was selected and transferred to a plate with 10% V8 juice agar and stored at 12 °C in the dark. Isolation of P. sojae from diseased plants was conducted following Dorrance et al. (2008). Plant tissue was washed to remove soil particles and then rinsed with sterile distilled water, and blotted dry on sterile filter paper. The sections from the edge of the lesion were placed on Phytophthora selective agar media. After 3 days of incubation, hyphal tips of Phytophthora were removed aseptically and transferred to fresh V8 selective plates.

Pathotypes characterization of P. sojae

The virulence formula of P. sojae isolates was performed by hypocotyl inoculation technique (Dorrance et al. 2008). A set of 14 soybean differential lines including ‘Harlon’ (Rps1a), ‘Harosoy 13XX’ (Rps1b), ‘Williams 79’ (Rps1c), ‘PI103091’ (Rps1d), ‘Williams 82’ (Rps1k), ‘L76–1988’ (Rps2), ‘Chapman’ (Rps3a), ‘PRX146–36’ (Rps3b), ‘PRX 145–48’ (Rps3c), ‘L85–2352’ (Rps4), ‘L85–3059’ (Rps5), ‘Harosoy 62XX’ (Rps6), ‘Harosoy’ (Rps7), and ‘PI 399073’ (Rps8) (McBlain et al. 1991; Dorrance et al. 2004) were inoculated, and Williams (rps) was used as a susceptible control. Ten seeds of each differential line were grown in vermiculite in the greenhouse at 25 °C, with a photoperiod of 14 h light/ 10 h dark for 7 days. Make a slit about 3 mm long with the scalpel in the hypocotyls of the seedling 1 cm below the cotyledonary node. The mycelium culture was cut in 3 mm square strips and placed the strips into the slit. Moisturized the plants in a box for about 12 h to prevent the agar from drying and for the infection to take place in the greenhouse. Susceptible plants would die or develop distinct lesions 3 to 5 days after inoculation. Seven or more plants killed was susceptible, three or less killed was resistant, and four to six plants killed was intermediate resistant. Every assay was repeated twice.

DNA extraction, amplification, and sequence analysis

Genomic DNA was extracted and purified using a DNAsecure Plant Kit (Tiangen Biotech, Beijing, Co., Ltd., China) following the manufacturer’s instructions, and was then stored at − 20 °C. Avr1c alleles were amplified in a 25 μL PCR reaction mix containing 2.5 μL of 10× PCR buffer (Mg⁺; TaKaRa, Japan), 2 μL of 2.5 mM deoxynucleoside triphosphates (dNTP) (TaKaRa), 0.5 μL of each specific primer (10 μM), 1 μL (5 U/μL) of rTaq polymerase (TaKaRa), 30 ng of genomic DNA, and 18.75 μL of sterilized and deionized water. The reactions were performed at 95 °C for 5 min, followed by 32 cycles of 95 °C for 30 s, 60 °C for 30 s, and 72 °C for 1 min and a final extension at 72 °C for 10 min. The PCR products were resolved by ethidium bromide electrophoresis in 1.0% (w/v) agarose gels in 1× Tris-acetate-EDTA buffer, and the bands were sequenced. PsActin was used as the control.

Sample preparation, RNA extraction, and real-time quantitative reverse transcriptase-polymerase chain reaction

The roots were immersed in a zoospore suspension, and mixed root samples were collected at 0.5, 1.5, 3, 6, 9, and 12 hpi and frozen in liquid nitrogen. The samples were mixed and total RNA was extracted using a Total RNA Kit I (Omega, Norcross, GA) following the manufacturer’s instructions, and was stored at − 70 °C. The concentration of RNA was determined using a spectrophotometer (ND-1000; NanoDrop, Wilmington, DE) and diluted to 100 ng/μL using diethyl polycarbonate water (RNase- and DNase-free; Solarbio, Beijing, China). cDNA was synthesized using a PrimeScript™ RT Reagent Kit with gDNA Eraser (TaKaRa). Real-time quantitative reverse transcriptase-polymerase chain reaction (RT-qPCR) was conducted in a 20 μL reaction mix containing 20 ng of cDNA, 0.4 μL of primers (10 μM) specific for the target or reference gene (PsActin), 10 μL of SYBR Green Premix Ex Taq (Tli RNaseH Plus), 0.4 μL of ROX Reference Dye II (50×), and 6.8 μL of sterilized and deionized water. The primers used in this experiment were all confirmed by PCR from genomic DNA to ensure the specificity. RT-qPCR was conducted using a 7500/7500 Fast Real-Time PCR System (Applied Biosystems Inc., Foster City, CA) under the following conditions: 95 °C for 30 s followed by 40 cycles at 95 °C for 5 s, and 60 °C for 34 s; and dissociation at 95 °C for 15 s, 60 °C for 1 min, and 95 °C for 15 s. The RT-qPCR analyses were performed in triplicate.

Plasmid construction

Avr1c alleles and mutants without a signal peptide were cloned from the genomic DNA of different P. sojae isolates. For the expression in N. benthamiana, the amplified fragments of variants were ligated into pBinGFP2 (N-terminal tag green fluorescent protein (GFP) fusion). For the bombardment experiment, all the amplified fragments were ligated into pFF19 (Timmermans et al. 1990). All site-directed mutants were cloned from genomic DNA by overlapping PCR using mutated primers. Individual colonies of each construct were tested by PCR and verified by sequencing. The cloning primers are shown in Additional file 3: Table S3.

Agrobacterium tumefaciens infiltration, protein extraction and western blotting

N. benthamiana plants were grown in a greenhouse house under a 16 h 25 °C:8 h 22 °C, day: night regime. N. benthamiana plants used for infiltration were 5–6 weeks old. All divergent Avr1c variants and mutants fused GFP constructs were transformed into Agrobacterium tumefaciens strain GV3101 by electroporation. A. tumefaciens strains were cultured in Luria Bertani liquid medium for 12–16 h at 30 °C in a shaking incubator. The cultures were collected by centrifugation and resuspended in infiltration buffer (10 mM magnesium chloride, 10 mM MES, pH 5.6, and 150 mM acetosyringone) to an OD600=0.6, and then infiltrated into N. benthamiana leaves. Two days after agroinfiltration, agroinfiltrated leaves were frozen in liquid nitrogen and ground to powder using a mortar and pestle. Approximately 0.5 g powder was added to 1.0 mL ice-cold extraction buffer (NP-40 (P0013F, Beyotime, Shanghai); 1 mM Phenylmethanesulfonyl fluoride (PMSF)) in a 2.0 mL centrifuge tube. The mixture was centrifuged at 14000 × g for 10 min at 4 °C, and 100 μL of supernatant was transferred to a 1.5 mL tube and then boiled in SDS-PAGE loading buffer for 5 min. The samples were loaded onto a 15% SDS-PAGE gel for protein electrophoresis.

The total protein in leaves of N. benthamiana expressing Avr1c variants and mutants were examined by western blotting with anti-GFP antibody. The separated proteins were transferred from the gel to polyvinylidene difluoride (PVDF) membrane and then blocked with PBSTM buffer (phosphate-buffered saline with 0.1% Tween-20 and 5% non-fat dry milk) for 30–40 min at room temperature with 60–70 rpm shaking; anti-GFP (1: 5000; #M20004, Abmart, Shanghai, China) was then added to PBSTM buffer and incubated at room temperature for 2 h, followed by five washes with PBST buffer. The membrane was then incubated with a goat anti- mouse antibody (Odyssey, no. 926–32,210; Li-Cor, Lincoln, NE, USA) at a ratio of 1: 10000 in PBSTM at room temperature in the dark for 30–40 min with 60–70 rpm shaking. The membrane was washed five times with PBST, and then visualized by excitation at 800 nm. The green fluorescent protein (GFP) was used as a control.

Gene co-bombardment on soybean leaves

Avr1c alleles and mutations were amplified using a pair of specific primers (Additional file 3: Table S3) that do not encode a signal peptide, and were ligated into the plasmid pFF19 (replacing GUS), which was digested by SmaI (Kong et al. 2015). Double-barreled particle bombardment assays were performed on 12–13 days old soybean leaves containing GUS DNA and empty vector (EV), or GUS and Avr, using a Bio-Rad He/1000 Particle Delivery System (Hercules, CA) (Dou et al. 2008b). Next, the soybean leaves were moisturized for 2 days in darkness at 25 °C. The leaves were stained for ~ 12 h at 37 °C with 0.5 mg/mL X-gluc (5-bromo-4-chloro-3-indolyl-β-D-glucuronic acid, cyclohexylammonium salt), 80 mM sodium phosphate, 0.4 mM potassium ferricyanide, 0.4 mM potassium ferrocyanide, 10 mM disodium EDTA (pH 8.0), and 0.1% (v/v) Triton X-100, and then destained in 75% ethanol. Each experiment was repeated at least three times.

Blue spots were counted and the ratio of blue spots in leaves expressing target gene compared with EV were calculated. The bars represent standard errors from at least three independent replicates and analyzed by t-test.

Structure of Avr1c variants

The structure of Avr1c variants were simulated using SWISS-MODEL (Waterhouse et al. 2018) and the RXLR effector PcAVR3a4 was used as a template, which has 23.08% sequence identity to Avr1c (Yaeno et al. 2011). Because mature AVR3a4 without the signal peptide (Asn22-Tyr122) was highly soluble and properly folded, and the N-terminal region (including the RXLR domain [Asn22-Arg58]) was disordered, so we used Avr1c^K79-H126 (deleted signal peptide Met1-Ala17 and Ser18-Gly78) in the simulation. The peptide Avr1c-1^K79-H126, Avr1c-4^K79-H126 and Avr1c-8^K79-H126 as target sequences were used to search for templates and build model.